'Queen Of The Internet' Mary Meeker Calls AI The Most "Unprecedented" Tech Shift Yet

NDTV

3 days ago

New Delhi:
Venture capitalist Mary Meeker, once called the "Queen of the Internet," is back with her first big trends report since 2019. Titled " Trends - Artificial Intelligence," the 340-page deep dive argues the rise of AI is unlike any tech revolution seen before.
"The pace and scope of change related to the artificial intelligence technology evolution is indeed unprecedented..." Ms Meeker writes, using the word "unprecedented" 51 times in the report.
Rise Of AI
Ms Meeker's report talks about how quickly AI has scaled. Instagram, WhatsApp, and YouTube took 2-4 years to hit 100 million users. ChatGPT did it in under three months. By April this year, ChatGPT had 800 million weekly users and now handles over 365 billion searches annually.
Based on Morgan Stanley data, it took 6-12 years for half of US households to get access to mobile and desktop internet. For AI platforms, Ms Meeker predicts the same will happen in only three years.
India Has A Major AI User Base
India has turned out to be a vital market for AI platforms. The country contributes the highest percentage of mobile app users for ChatGPT (13.5 per cent), ahead of the US (8.9 per cent) and Germany (3 per cent). It is also the third-largest user base (6.9 per cent) for China's DeepSeek. "India has been a key user-base market for AI companies," the report said.
Open Source vs Closed Source AI
Ms Meeker says AI is splitting into two paths: closed models like GPT-4 and Claude, and open models like Llama and Mixtral.
Closed models lead in performance and are favoured by enterprises, but they lack transparency.
Open models are more accessible and are driving innovation in local languages, grassroots tools, and sovereign AI efforts.
"We're watching two philosophies unfold in parallel, freedom vs control, speed vs safety, openness vs optimization - each shaping not just how AI works, but who gets to wield it," Ms Meeker writes.
China, for instance, is leading the open-source race. As of Q2 2025, it has released major models like DeepSeek-R1, Alibaba's Qwen-32B, and Baidu's Ernie 4.5.
Falling Costs, Soaring Competition
While training costs for models have gone up (reaching up to $1 billion), inference costs have dropped by 99 per cent in two years, according to Stanford data. Nvidia's 2024 Blackwell GPU uses 105,000 times less energy per token compared to its 2014 predecessor. Google's TPU chips and Amazon's Trainium are also scaling rapidly. "These aren't side projects, they're foundational bets," Ms Meeker notes.
Financial Reality Check
Despite massive user growth, she has warned that revenue per user is still low with a median of $23 (Rs 2000) across most platforms. The industry is burning through cash. Investors are betting big, but it is still unclear who will come out on top. "Only time will tell which side of the money-making equation the current AI aspirants will land," she writes.
AI Already Shaping The Real World
AI is moving beyond apps, the report says. It is driving cars, running factory robots, and aiding in healthcare. Jobs aren't vanishing but evolving, with AI becoming a co-pilot for coders, writers, and analysts. AI-related job listings have jumped 448 per cent since 2018 Ms Meeker says.
The Infrastructure Race
Access to powerful tech, like GPUs, chips, and data centres, is now a global competition. Mary Meeker compares it to the space race during the Cold War.
There are serious concerns too. AI can be biased, spread wrong info, or behave unpredictably. Ms Meeker says we need clear rules, honest leadership, and smarter systems to handle AI's fast growth.